Treesearch
Mating design considerations—How many crosses do we really need to test?
| Authors: | R. Johnson |
| Year: | 2000 |
| Type: | Scientific Journal |
| Station: | Pacific Northwest Research Station |
| Source: | In: Proceedings 25th biennial southern forest tree improvement conference. SFTIC. 47: 93-100 |
Download (PDF 681 KB): https://research.fs.usda.gov/download/treesearch/4923.pdf
Abstract
The impact of increasing the number of crosses per parent on the efficiency of backwards selection was examined using Monte Carlo simulation. Both the efficiency of reselection and its associated variance leveled off after two to three crosses per parent.Because so few crosses appear to be needed to estimate parental GCA values, a quasi-complementary mating design option was investigated where only a subset of the full-sib families were planted into replicated field trials, while the remainder were established in less expensive pure-family full-sib blocks. Breeding values for families planted only in blocks were generated from the family data available from the replicated trials. Compared to using only the field tested families, significant additional gains could be made through the increased selection intensity resulting from the additional families planted only in blocks.